Ok, so finally getting to this! Sorry it took a bit. I didn’t want to draw attention away from the Get Large Jam and then September and October was a bit busier for me than I expected so I am quite a bit behind.
With that out of the way we will start with our first post about the voting irregularities that happened last year. A quick thing to note is I will make lots of references to things like TL0, TL1, ect, ect. These are trust levels that are assigned in discourse based on user activity. All you really need to know is that TL0s are new users while everything else above it are active users.
Also, there will be a lot of talking back and forth between WasteLine and RogueWeight. I am not trying to say one is better then another (I personally have not had time to play either of them), but more pointing out that the wave of voting RogueWeight experienced was mainly being motivated by the fact it did not have AI generated content instead of its quality as a game. The time between the calls to vote and the votes is so short its highly likely most of the TL0 votes they received did not play, much less even go to the games itch page before voting.
TLDR;
We received reports from another user that WasteLine was trying to cheat the community vote. After investigation we did not find that WasteLine was cheating the vote in any manner. This was not the answer the reporting user wanted as they just really wanted them disqualified due to their use of AI generated content. This lead to a smear campaign against Jupiter in order to try to put pressure on us to disqualify them and when that was not working they resorted to doing the same things they were accusing WasteLine for to boost the second place game (RogueWeight) to first place to ensure WasteLine would not win community choice.
The main point I would like to get out there is that we did not find any evidence that Jupiter attempted to rig the community vote in any manner. In fact we found evidence to the contrary, with Jupiter discouraging any cheating even when things got rough and instead seeing a 3rd party working to try to get WasteLine disqualified and make sure they lost the vote just because they made use of AI generated content.
While this 3rd party skirted the rules, the way they pushed votes towards RogueWeight was clearly against the spirit of the vote, and the way they tried to smear the reputation of another dev in the community without any evidence was frankly sickening. This was a clear case of a hand full of people who didn’t even participate in the jam throwing their weight around and doing whatever they can to try to change a result they did not like.
How it Started
A few days before we were planning on closing the community vote we received a message from another user that said they believed that Wasteline was cheating by asking their community members to pad the likes on their post. We take claims like this very seriously so I asked for them to provide us with the evidence they had and that we would look into it asap.
The reporting user then provided us with a screen shot of a discord conversation between two users talking about creating dup accounts to pad the vote. Oddly enough though in the same conversation sent to us is the lead of Wasteline (Jupiter) telling the users not to do that. Either way though we thanked the user for the info and told them we would look into it immediately.
Our Initial Investigation
We quickly reached out to Jupiter and began gathering info looking for any possible cheating. What we found out is that there were two users in Wasteline’s discord that were talking about padding the votes (the same two from the screenshot) but Jupiter thought the warning was sufficient and that they were joking so didn’t think they should bother us with it.
Looking at the context of the messages all 3 of us agreed that we would have come to a similar conclusion in their shoes and felt it was a reasonable explanation. Along with that, looking over our site’s data we did not find anything that showed any significant vote padding was going on. We did ask Jupiter to help us check if those two users did double vote so we could remove it if they did and Jupiter agreed to help.
We then got back with the results to the reporting user. We told them that we found no evidence that Wasteline’s team was encouraging any cheating, but we did find two users in their discord may have been cheating. Due to this there is no evidence that justifies disqualifying Wasteline but we are working with them to identify and remove the duplicate votes from those two users.
This was unacceptable to the reporting user. They demanded that we disqualify Wasteline as they were obviously cheating.I explained again that we did not find anything that suggested the Wasteline team was cheating and that it looked like it was just two rogue fans doing something they shouldn’t.
The reporting user was adamant though that they were cheating, so we asked if they had any other evidence or reasons for believing this.They did not have any more evidence but did give the following reasons:
- Wasteline makes use of AI generated images and thus should be disqualified
- There is no way a game using AI generated content would win community vote so they must be cheating
- Wasteline is a trash game and would no way have that many votes legitimately
It was with this we saw that the users actual intention was to just get Wasteline disqualified because they used AI generated content. We thanked them for the tip, and said we were going to track down and remove any fraudulent votes we found and to let us know if they find any more evidence.
With that I thought that this matter was sorted, but I was wrong.
Continued Attempts at getting WasteLine Disqualified
Over the remainder of the incident we would continue to get reports from other people of Wasteline cheating, but all of these reports were just the same screen shot with similar justifications. Talking with some of these people it was obvious they were not told the whole truth as they backed off quickly after having the situation explained. A larger majority was obviously being told to spam report them to try to put pressure on us to disqualify Wasteline.
Along with this a new justification was being thrown about by both the original user and these new people. They claimed that Wasteline should be disqualified because it was Jupiter’s community that was deciding the vote, and not Weight Gamings.
This reasoning, at least to us, was a stretch to say the least. We could see from the numbers we had that the majority of WasteLine’s votes were spread fairly well between the TL2 and TL1 users. This told us that the reverse was actually more likely, most of their community are Weight Gaming users. They did have a fair chunk of TL0 users as well, but nothing beyond what we would expect and nothing that would have really changed the result all that much at that point.
Also, while the concern on its face is fair this has always been a problem when it comes down to popularity votes like this. In the past, if a dev had a large following outside of the site it was not hard for them to win the community vote just by leveraging it. When pressed on this point we received much of the same expected answer. Those instances are different because they didn’t use AI generated content.
The Vote Spikes
It was late at night and I was trying to get to sleep since I had to be up early for work the next day when my phone buzzed. Normally I would ignore it, but I knew we were close to finishing up so I figured Alex may have had a question I didn’t want to keep hanging.
Instead, Alex was letting me know that there was a sudden surge of voting for Rogue Weight. This was odd and a red flag as their voting has been stable for almost a month up to that point, so a sudden jump like that is very suspicious. Alex wanted to call it right there and then for WasteLine as it was obvious something was up, but it was late for me and I wanted to confirm that there was no funny business before we jumped to any action. I told Alex that we should just extend voting by 3 days to give WasteLine a chance to respond and give me time to review the votes for any issues.
For me personally, this was a mistake and in hindsight I should have gone with Alex’s suggestion on the matter.
Looking over the votes for RogueWeight it was obvious something was going on. Before this spike RogueWeight had no likes from TL0 users with all of their likes spread mainly between TL1 and TL2 users.But now they suddenly jumped out to almost a quarter of their votes coming from TL0 users. When we looked at a timeline of when the likes came in we could also see a clear pattern where almost all of them were coming in around the same time.
With this information we were able to map all the spikes back to social media posts and streams being done by the same people trying to get WasteLine disqualified. Looking over them we can see them urging people to join just to vote for RogueWeight and was framing it as a “fight against AI”.
While this is not explicitly against the rules it was against the spirit of the vote since the motivating factor for most of them was obviously not the quality of RogueWeight, but the fact that WasteLine used AI generated content. Further compounding it these same people were trying to use the jump in votes to further push disqualifying WasteLine by saying its proof that WasteLine votes were invalid was the fact that RogueWeight was now in first.
This personally was a very sore point for me, and I could not tell if they were so dense as to think we would not be able to figure out what they were doing or simply just did not care that they were doing what they were accusing WasteLine of.
What we Decided to Do
Unfortunately we were between a rock and a hard place. We never thought a proxy war like this would break out in the community vote. Also, at the end of the day no rules where being broken and while we saw clear evidence of at least the group trying to get WasteLine disqualified was pulling some shady stuff the dev of RogueWeight had nothing to do with it and we would feel bad about punishing them for another group’s actions.
Due to this we decided to let it play out. This resulted in a very aggressive back and forth “tit for tat” voting between WasteLine and the group trying to make sure they did not win community vote. Finally, when the time came for us to close voting we saw that RogueWeight won by a few votes so decided to call it for them.
Final Thoughts and What We can do In the Future
At the end of the day, when all is said and done we only found 3 fraudulent votes with the help of Jupiter and the WasteLine team. And the only reason our anti-spam system did not catch it is all 3 accounts were already existing accounts that the users in question abandoned years ago. And over all that was ~1% of WasteLines votes, well within what we would expect for most games since we always expect some cheating, usually from people just trying to show support for their favorite creators.
So, was all this justified over 3 votes? In my mind the answer is no, but the cheating was not the point for most of the people trying to push for WasteLine to be disqualified. It was more the fact that in their minds there was no way a game that even made partial use of AI generated content could win community choice. The more I interacted with them the more I began to believe that their real fear was that this “unified front” against AI may not be as unified as they may have believed.
I think sometimes a lot of people think I kind of live in my own little world but in reality I seek out opinions from a wide array of people and I am in quite a few communities where I just sit and observe what is going on. It’s one of the reasons I know more about what was going behind the scenes as I could see the conversations being had in other communities and discords where they just forgot I was there. I’ll be getting more into it in my AI discussion but what I have learned from my interactions is the general community’s thoughts on AI are not as clear cut as we may want to believe. I have even seen artists in the same circle have completely different opinions on the tech. I have also seen closed servers completely turn on their own members, their own friends, who dared to use AI to create placeholders for a demo and then ask for help replacing those placeholders with actual art.
Can you just imagine that, being persecuted and told you are not wanted in the very same space you were asking for help, by those who you thought were your friends for one misstep. Do you imagine that helped anyone at the end of the day?
I digress though. To get back on the topic at hand, how can we deal with something like this in the future? Tbh, I don’t quite know yet. There are a few options that we are considering, some suggested by others that we think are worth pursuing but not sure.
- Restricts likes to TL1 and up during the jam
- This would help with the tit for tat voting by requiring users to have to spend some time in the forums to vote but I am not sure if we can limit it to specific categories so it would likely have to be a site wide setting
- Find a different way to handle voting
- About the same as the above but use an external voting system for community vote. I am not sure what we can use for this and we may have to roll our own. This may not be feasible to do with the forums though so may only be viable with the main site.
- Rework the rules to allow us to handle suspicious voting better
- It would be nice to allow us to handle suspicious voting a bit more at our discretion, but it can be hard to separate legit from illegitimate votes outside of obvious situations like what happened last year and could feel like we are tipping the scales when we enforce it.
- Get rid of community voting
- As it says on the tin, we get rid of community voting all together.
At the moment we are still mulling it all over so we are not fully sure what we will do yet. Though I do think the suggestion of requiring TL1 to like posts would be a fair restriction that would be easy enough to implement.